Very slow loading of the Convnet pretrained model on lesson 1

[EricPB (Eric Perbos-Brinck)]

You’re not alone

Lesson 2 notebook = memory hog?

I’m testing the Lesson 2 Notebook on Amazon Rainforest on my personal DL rig (Ubuntu 16.02 + GTX 1080Ti + 16Gb RAM and 50Gb Swap), using GNOME System Monitor and PSensor. It crashed repeatedly before, getting “OOM” (Out Of Memory) errors around 3/4 of the work. It seems to keep all the previous cell runs in memory, like here I’m on Cell 24 with sz=128 learn.set_data(get_data(sz)) learn.freeze() learn.fit(lr, 3, cycle_len=1, cycle_mult=2) and my system, which I rebooted prior to launching…

I’ll try and look deeper into it tomorrow, following your tips.